X-Git-Url: http://git.shiar.nl/sheet.git/blobdiff_plain/2f6bd0f228e04949e561c9e5d7713a0baef77cd5..1303498b614eb9ca37bf8a9410776636cc837ba0:/charset.plp

diff --git a/charset.plp b/charset.plp
index ed495fe..726ca7c 100644
--- a/charset.plp
+++ b/charset.plp
@@ -36,18 +36,10 @@ else {
 	say "Charset comparison:";
 }
 
-sub optionlink {
-	my ($title, $href, $selected) = @_;
-	return sprintf(
-		$selected ? '<strong>%s</strong>' : '<a href="%2$s">%s</a>',
-		EscapeHTML($title), $href
-	);
-}
-
 print join " •\n", (
 	map {
 		join " ·\n", pairmap {
-			optionlink($b || ucfirst $a, '/charset'.($a && "/$a?compare"), $a eq $Request);
+			showlink($b || ucfirst $a, '/charset'.($a && "/$a?compare"), $a eq $Request);
 		} @{$_}
 	}
 	[
@@ -147,12 +139,14 @@ sub tabinput {
 		eval { $charset->{setup}->(\%row) }
 			or Alert("Incomplete setup of $input", $@);
 	}
+	$row{endpoint} ||= 0xFF;
 
-	if ($row{set}) {}
+	if (defined $row{table} or defined $row{cell}) {
+		$row{set} //= $input;
+	}
 	elsif ($row{set} = Encode::resolve_alias($input)) {
 		$row{offset} = delete $row{startpoint};
-		$row{endpoint} ||= 0xFF;
-		if ($row{set} eq 'MacHebrew' or $row{set} eq 'MacThai') {
+		if ($charset->{varchar}) {
 			# array of possibly multiple characters per code point
 			$row{table} = [
 				map { Encode::decode($row{set}, pack 'C*', $_) } $row{offset} .. $row{endpoint}
@@ -164,43 +158,47 @@ sub tabinput {
 			$row{table} = Encode::decode($row{set}, pack 'C*', $row{offset} .. $row{endpoint});
 		}
 
-		if ($row{set} eq 'cp437') {
-			if ($row{offset} <= 0xED and $row{endpoint} >= 0xED) {
-				# replace phi glyph
-				substr($row{table}, 0xED - $row{offset}, 1) = 'ϕ';
-			}
-			if ($row{offset} < 0x20) {
-				# replace control characters by visible variants
-				my $sub = substr ' ☺☻♥♦♣♠•◘○◙♂♀♪♫☼►◄↕‼¶§▬↨↑↓→←∟↔▲▼', $row{offset};
-				substr($row{table}, 0, length $sub) = $sub;
-			}
-		}
-		elsif ($row{set} eq 'symbol') {
-			if ($row{offset} <= 0x60 and $row{endpoint} >= 0x60) {
-				# replace radical extender by closest unicode equivalent
-				substr($row{table}, 0x60 - $row{offset},  1) = '│';
-			}
-			if ($row{offset} <= 0xBD and $row{endpoint} >= 0xFF) {
-				substr($row{table}, 0xBD - $row{offset},  2) = '⏐⎯'; # arrow extenders
-				substr($row{table}, 0xD2 - $row{offset},  3) = '®©™'; # serif variants
-				substr($row{table}, 0xE0 - $row{offset},  1) = '◊'; # replace lookalike, should match AdobeSymbol
-				substr($row{table}, 0xE2 - $row{offset},  3) = '®©™'; # sans-serif variants
-				substr($row{table}, 0xE6 - $row{offset}, 10) = '⎛⎜⎝⎡⎢⎣⎧⎨⎩⎪';
-				substr($row{table}, 0xF0 - $row{offset},  1) = '€';
-				substr($row{table}, 0xF4 - $row{offset}, 11) = '⎮⌡⎞⎟⎠⎤⎥⎦⎫⎬⎭';
-			}
-		}
-
 		$row{endpoint} -= $row{offset};
-
-		$visible->{ascii} =  # assume common base
-		$visible->{ $row{set} } = 1;
+		$visible->{ascii}++;  # assume common base
 	}
 	else {
 		Alert("Encoding <q>$input</q> unknown");
 		return;
 	}
+
+	if (my $replace = $charset->{replace}) {
+		while (my ($offset, $sub) = each %{$replace}) {
+			$offset -= $row{offset};
+
+			if (ref $row{table} eq 'ARRAY') {
+				$row{table}->[$offset] = $sub
+					if $offset >= 0 and $offset <= $row{endpoint};
+				next;
+			}
+
+			my $length = length $sub;
+
+			if ($offset < 0) {
+				$offset > -$length or next; # at least one character after start
+				# trim leftmost part to start at offset
+				substr($sub, 0, -$offset) = '';
+				$length += $offset;
+				$offset = 0;
+			}
+
+			if ((my $excess = $row{endpoint} - $offset - $length + 1) < 0) {
+				$excess > -$length or next;
+				# trim rightmost part to prevent overflow
+				substr($sub, $excess) = '';
+				$length += $excess;
+			}
+
+			substr($row{table}, $offset, $length) = $sub;
+		}
+	}
+
 	push @request, \%row;
+	$visible->{ $row{set} } = 1 if $row{table};
 }
 tabinput($_) for @tablist;
 
@@ -324,7 +322,7 @@ for my $row (@request) {
 			my $cp = $offset + $row->{offset};
 			my $glyph = ref $row->{table} eq 'ARRAY' ? $row->{table}->[$offset] :
 				substr $row->{table}, $offset, 1;
-			my ($cell, $name, $class) = $glyph eq $NOCHAR ? () :
+			my ($cell, $name, $class) = !defined $glyph || $glyph eq $NOCHAR ? () :
 				$glyphs->glyph_html($glyph);
 
 			if ($mode) {